You might imagine that artificial intelligence is globalized, and yet it’s deeply “culturalized.”
We questioned the American model ChatGPT, the French Le Chat, and the Chinese DeepSeek to evaluate a composition of objects from an artistic perspective and get advice on how to improve it:
The American model provided practical and precise recommendations on the distances between objects, in terms of both length and depth. The French model began with principles of decoration (an American user might wonder “What’s the point?”), and we had to insist to get concrete specifics from it.
This perfectly illustrates the “gap between pragmatism and conceptualization” described in our intercultural model: Americans tend towards pragmatism (“Just do It”), while the French lean towards conceptualization (“I think, therefore I am”).
We had some fun asking both models if this difference could be explained:
The French model posited that “the different artificial intelligence models can indeed have different approaches and styles depending on their design, training, and the data they were trained on.” Meanwhile, the American model immediately listed the editorial differences between ChatGPT and Le Chat. So, the French model shared an intellectual overview, and the American, its practical application.
There are other, equally cultural, differences:
- The American model tends to use informal “you” (tutoyer), while the French model uses the formal “you” (vouvoyer).
- When a strong dissatisfaction is expressed, the American model says you’re right and I will try to do better, while the French model offers a technical problem as an excuse to stop the conversation. It’s hard not to connect this to American “good citizenship” and the French “sense of honor.”
- As for the Chinese model DeepSeek, it suggested, among other things, “balancing the number of objects per culture” to promote “East-West dialogue,” plunging us right into geopolitics!
Two conclusions from this:
- Intercultural intelligence is needed to enhance artificial intelligence.
- There are no good or bad models. There are complementary models.
This diversity is an opportunity for those who know how to leverage it.
On the GapsMoov platform, see the gaps:
- Between pragmatism and conceptualization.
- Between mission and obligation.
- Between conflict and partnership.
